Preview in some words a project funded by the European Commission

Eurorisk PREVIEW Project overview

PREVIEW in some words

• A project funded by the European Commission

• 23 millions € of eligible costs, 14 millions € granted

• New information services to help risk management

• Gathering users needs
• Using the most advanced research and technology
• Validation on operational platforms

• 58 partners from 15 nations

• Scientists
• Operators
• Industrial companies
• End Users

• 45 months performance schedule and yearly budget reviews

Cost figures

Atmospheric Risks services

• FIRES

Geophysical risk services

Man made Risk services - Engineering - Industrial accidents

Work Breakdown Structure

Components of Package

Risk Mapping

• Lead:

• SMHI

• Task:

• Map return periods
• Estimate climatology

Best Practise Probability Forecasts

• Lead:

• Met Office

• Contributors:

• ECMWF (passive)
• Meteo-France
• DWD
• Met.no
• ARPA-SIM

Best Practise Probability Forecasts

• Ensemble Forecast inputs:

• Medium-Range (3-10 days)
• ECMWF
• ARPA-SIM (COSMO-LEPS) (Days 3-5)
• Short-Range (1-2 days) – multi-model ensemble consisting of contributions from:
• Meteo-France - PEACE
• DWD - SRNWP-PEPS
• Met.no - TEPS/LAM EPS
• Met Office - LAMEPS/EPS

• Post-processing…

Site-specific Ensemble Forecasts

• Ensemble forecasts will be collected and post-processed on a site-specific basis:

• Utilise existing technology/capability
• Allows bias correction and calibration
• Reduced data volumes for international exchange
• Aids combination of multiple forecast inputs in common format

• Disadvantage:

• Risk of strongest winds falling between site locations
• Hi-resolution models and nowcasts add detail in short-range

Site-specific post-processing - Kalman Filter MOS

• Kalman Filter MOS:

• Statistical model which relates model fields to observed windspeed at the site.

• Main features:

• Corrects site biases

• 60-day training cycle allows rapid adjustment for model changes

• Available for any site worldwide with observations

• Can be set up for each model to correct its own biases

Calibration of Probability Forecasts

• Calibration forecasts of a single “event” is straightforward using a reliability diagram:

• 70% EPS prob50% issued

Calibration with rank histograms

• Bin relative frequencies give probabilistic weights.

• No threshold dependence (flexible)

• But weights vary with:

• season
• parameter
• forecast range/time of day
• location (reduced by KF)

Calibrated probability distribution functions...

• More information about ‘extremes’ plus increased reliability.

EPS Meteogram

• Ensemble spread and forecast trends

• Box shows 25-75% range
• Whiskers show full range (or 95% after calibration)
• Central bar shows median

• Other models can be added

Products for the Risk Manager

• Plot of ensemble spread

Verification

• Site-specific

Additional short-range forecasts

• Met.no - hi-res downscaling (complex terrain)

• Met Office – hi-res downscaling

• Met Office – wind nowcasting

Warnings/Downscaling

EMMA Website - for display of warnings

Training & Awareness

• Workshop topics:

• Uncertainty
• Best practice forecasting
• Risk management
• How to get best value for specific users

Civil Protection Response

• Demonstrate warnings and (potential) responses

• Assess impact compared to traditional methods

Tasks Lead Contributors

Some more detail

Summary

Issues

• SRNWP – PEPS – should we have a separate KF-MOS for each member?

• Downscaling to local impacts – how? SRSA?

• How will risk maps be used in warning process? (Ken unclear)

Accreditation